AI songwriting tools come in many shapes and sizes.
This article focuses on AI songwriting tools for musicians, defining the most common workflows and sharing tips about the most popular apps.
Most musicians don't want to automate away their creativity. If they're using AI tools, it's to help break through steps in the process where they tend to get stuck.
Table of contents
What is AI Songwriting?
The expression AI songwriting refers to creative workflows where artificial intelligence augments human songwriting. That can include AI lyric generation (like ChatGPT) as well as music generation (like Suno).
It's helpful to distinguish songwriting from beat making. Songs consist of meta-categories like the verse, bridge and chorus. They center on chords, melodies, and lyrics more than rhythm sections and drum loops.
Every musician approaches songwriting in a unique way. You'll need to explore the available tools to decide what your stack should look like. Start by figuring out where you're stuck. Let that be your entry point into the AI product landscape.
Songwriting AI tool categories
We organized songwriting AI tools into a few big categories, according to the problems that they solve for musicians. Our website features free, full length articles for each category and we've linked to them here for your convenience.
AI chord and melody generators are used by musicians to come up with raw musical material from scratch. AI MIDI generation is the most direct solution for most songwriters, but instrumental audio formats are also valid.
AI lyric generators are able to ingest lyrical themes from a user and transform them into full, rhyming schemes.
AI song generators combine lyric, chord and melody generation into a single app. They tend to be based on a lyric-to-song experience, coupled with genre, mood or style prompting. Some apps, like Suno and Riffusion, even have image-to-song.
AI music extenders ingest audio from a user and expand on it in a number of ways, either by writing new sections or composing arrangements around existing music.
Device type is a second major factor to consider, as each comes with its pros and cons. Mobile and tablet applications may be more convenient for songwriters working at a guitar or piano, while desktop apps and DAW plugins are usually better for composers working in the studio.
Now that we've identified the product landscape, let's have a look a the hands-on workflows that music producers have been experimenting with.
AI-Powered MIDI Songwriting Tools
Some songwriters prefer to work directly on the chord charts and melodies of an existing song. If that's the case, we recommend checking out our article on HookTheory's new AI MIDI extension feature, Aria.
Aria is a state of the art AI MIDI model developed in partnership with a professor at Carnegie Melon University and part-time Google DeepMind researcher. They fine tuned it on HookTheory's database of 50,000+ MIDI song transcriptions.
AI Songwriting with Hookpad's Aria feature
Users input their chord chart and melody quickly in Hookpad.
Select empty measures following an existing section and prompt Aria to compose new MIDI chords and melodies in that region.
Select measures with existing MIDI content. Aria will study the musical context before and after to suggest new possibilities.
Our full length article on AI MIDI Generators covers this topic in more detail but we hope this has provided you with a general idea of what's possible.
If you're generating AI MIDI ideas and applying sound design in a DAW, consider using AudioCipher's MIDI Vault to keep those ideas organize and accessible.
Organizing song ideas in a MIDI file manager
How to organize your song sketches in AudioCipher
At the end of each songwriting session, save your best ideas in AudioCipher to make sure they're easy to retrieve in the future. Here's the step by step process.
Save the session's most important audio and MIDI files to your hard drive
Open the MIDI Vault as shown in the video above and create a new card
Drag and drop the audio files from your harddrive to the card
Add a card title and metadata for the bpm, key, instrument type, mood, genres and any long form notes that may be important in the future
During later sessions, you'll be able to filter through the vault and find that card with all the song ideas in tact. Drag them from AudioCipher into a new DAW project and pick up where you left off.
This workflow is considerably better than old school nested-folder structures. You no longer have to resort to stuffing metadata into your file names or bury ideas in DAW projects, never to be opened again.
Visit the MIDI Vault website to learn more and watch some video demos.
Writing songs with AI lyric generators
Your average songwriter hammers away at chords and mumbles half-words with melodies until something emerges. AI lyric generators take a more direct path to the end goal, consuming initial ideas and turning them into song lyrics.
Lyrical style is the number one challenge with generative AI. It can generate text on any theme, but when you request lyrics it almost always defaults to four-bar end rhyme schemes. That's not how most lyricists actually write today.
The best solution is to use a high quality language model, like ChatGPT or Claude, and giving it some examples of the lyrical style you want to imitate.
Prompting tip for retrieving AI lyrics in your desired style
Use this prompt with an LLM like ChatGPT to get lyrics in the style of your choice. Replace the bracketed sections like an ad lib, to customize the input: "You are a [music style] lyricist working on a new song. You do not want to write in standard four-bar end rhymes schemes. Instead you will be writing lyrics that reflect a specific style. I will share three examples with you below. Read each of them and study them closely to understand the structural similarities. Focus on details like rhyme schemes (notice where they do or do not occur), along with phrase length, tone of voice, word choice, and other stylistic nuances. Example 1: [insert full lyric of a song] Example 2: [insert full lyric of a song]
Example 3: [insert full lyric of a song]
Confirm that you're ready and I will provide a lyrical theme for you to write about, in this style."
You can input lyrics from other songs you've written to see how close the LLM gets to understanding your writing style.
Expanding on existing song ideas with AI
AI music extension is our favorite technique in the ecosystem today. There are only a few companies who do it well, so we'll list them here along with some suggestions on how you can use them in a songwriting workflow.
Music can be extended horizontally (time) and vertically (accompaniment):
Completing an unfinished song section
Generating new song sections to come before or after an existing section
Hearing possible instrumental arrangements for chord and melody
Exploring lead melodies in the same key and BPM as your progression
Audio-to-Audio extension with Suno and Udio
The two best audio to audio song extension apps are Udio and Suno. They each bring something unique to the table and serve different functions. On both platforms, the audio output is free to download and use commercially.
Suno excels at re-arranging chord and melody compositions that were performed on a single instrument.
Users upload a solo instrument recording and combine it with a text prompt to guide the mood, style or genre.
Users upload the full song idea and prompt for a similar genre. The extension won't sound identical to the audio input, but you may very well get some interesting ideas from it.
Udio excels at maintaining an existing audio track and writing more music in that same style.
Users upload an audio file, be it solo instrument or full arrangement, and Udio will done of two things; either it continues writing in that style or it will sample the reference track with a new instrumental accompaniment.
Download the full track or individual stems from Udio.
These tools can be used in tandem to develop existing audio recordings. You can take this a step further with an AI audio-to-MIDI converter like SampLab or RipX.